一种基于细节特征的MCA多聚焦图像融合方法  

Multi-focus Image Fusion Method Based on Detail Feature in Morphological Component Analysis

作　　者：陈杰[1] 陈荣鑫[1] 谢书童[1] 

机构地区：[1]集美大学计算机工程学院,福建厦门361021

出　　处：《电视技术》2015年第15期1-5,27,共6页Video Engineering

基　　金：国家自然科学基金项目(61100106);福建省自然科学基金项目(2014J01245);集美大学科研基金项目(ZQ2014003)

摘　　要：针对多聚焦图像的特点,提出一种形态分量分析分解下的基于细节特征的图像融合方法。该方法首先将原图像分解成光滑部分和纹理部分。其次,为了充分体现多聚焦的纹理特征,在光滑部分采用微分算子定义的细节特征作为融合规则。最后,在纹理部分采用选择绝对值作为纹理部分的活跃程度的度量,使用绝对值最大法作为其融合规则。实验结果表明,该算法得到的融合图像优于其他多聚焦融合的算法。According to the characteristic of multi-focus images, a fusion method based on detail feature is presented in morphologi- cal component analysis. In this method, source images are separated into cartoon and texture components firstly. In order to em- body the feature of the multi-focus image fully, the cartoon components are fused with a rule of the detail feature which is defined by using differential operators secondly. Finally, The absolute value is used to measure the activity level of the texture components, so the texture components are fused with a rule of maximum absolute value. The experimental results show that the proposed algo- rithm outperforms other approaches.

关 键 词：图像融合 形态分量分析 光滑部分 纹理部分 细节特征 微分算子 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]